gru_test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 311,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch import nn\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = np.arange(0, 10, .1)\n",
    "y = np.sin(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD8CAYAAABzTgP2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(x, y)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 949,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[ 0.0298, -0.0699,  0.0535, -0.0190,  0.0137],\n",
       "         [ 0.0486, -0.1176,  0.0884, -0.0471,  0.0293],\n",
       "         [ 0.0673, -0.1648,  0.1229, -0.0748,  0.0447],\n",
       "         [ 0.0856, -0.2114,  0.1569, -0.1020,  0.0598],\n",
       "         [ 0.1036, -0.2572,  0.1901, -0.1286,  0.0746],\n",
       "         [ 0.1212, -0.3019,  0.2224, -0.1545,  0.0890],\n",
       "         [ 0.1382, -0.3456,  0.2538, -0.1794,  0.1030],\n",
       "         [ 0.1546, -0.3880,  0.2840, -0.2033,  0.1164],\n",
       "         [ 0.1705, -0.4290,  0.3130, -0.2261,  0.1294],\n",
       "         [ 0.1858, -0.4685,  0.3407, -0.2478,  0.1417]],\n",
       "        grad_fn=<ThAddBackward>),\n",
       " tensor([[ 0.0298,  0.0523, -0.0812, -0.0771,  0.0277],\n",
       "         [ 0.0723,  0.0548, -0.1144, -0.1086,  0.0268],\n",
       "         [ 0.1142,  0.0574, -0.1474, -0.1398,  0.0260],\n",
       "         [ 0.1552,  0.0601, -0.1800, -0.1706,  0.0253],\n",
       "         [ 0.1951,  0.0630, -0.2122, -0.2010,  0.0245],\n",
       "         [ 0.2336,  0.0661, -0.2439, -0.2308,  0.0239],\n",
       "         [ 0.2706,  0.0695, -0.2751, -0.2600,  0.0233],\n",
       "         [ 0.3060,  0.0731, -0.3058, -0.2885,  0.0229],\n",
       "         [ 0.3395,  0.0772, -0.3358, -0.3163,  0.0225],\n",
       "         [ 0.3711,  0.0816, -0.3651, -0.3432,  0.0223]],\n",
       "        grad_fn=<TanhBackward>))"
      ]
     },
     "execution_count": 949,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class MyRNN(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, nonlinearity=torch.tanh, **kwargs):\n",
    "        super().__init__()\n",
    "        self.hidden_size = hidden_size\n",
    "        self.w_hh = self.xavier_init([hidden_size, hidden_size])\n",
    "        self.b_ih = nn.Parameter(torch.zeros([1, hidden_size], requires_grad=True))\n",
    "        \n",
    "        self.w_ih = self.xavier_init([input_size, hidden_size])\n",
    "        self.V = self.xavier_init([hidden_size, hidden_size])\n",
    "        self.b_hh = nn.Parameter(torch.zeros([1, hidden_size], requires_grad=True))\n",
    "        self.nonlinearity = nonlinearity\n",
    "    \n",
    "    def forward(self, x):\n",
    "        seq_size = x.size(0)\n",
    "        self.h = torch.zeros([1, self.hidden_size], requires_grad=True)\n",
    "        for i in range(seq_size):\n",
    "            self.a = self.h@self.w_hh + x[i]*self.w_ih + self.b_ih\n",
    "            self.h = self.nonlinearity(self.a)\n",
    "            self.o = self.b_hh + self.h@self.V\n",
    "        return self.o, self.h\n",
    "    \n",
    "    def xavier_init(self, size):\n",
    "        return torch.nn.Parameter(torch.autograd.Variable(\n",
    "            nn.init.xavier_normal_(torch.empty(*size)),\n",
    "            requires_grad=True\n",
    "        ))\n",
    "n = MyRNN(1, 5)\n",
    "n(data[0:10, :].permute(1, 0).view(sq, -1, 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 950,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[ 0.0094,  0.0145, -0.1599, -0.0854, -0.0201],\n",
       "         [ 0.0133,  0.0313, -0.2304, -0.1159, -0.0331],\n",
       "         [ 0.0167,  0.0461, -0.2958, -0.1387, -0.0432],\n",
       "         [ 0.0197,  0.0590, -0.3548, -0.1545, -0.0502],\n",
       "         [ 0.0227,  0.0700, -0.4071, -0.1648, -0.0540],\n",
       "         [ 0.0258,  0.0791, -0.4529, -0.1710, -0.0551],\n",
       "         [ 0.0289,  0.0863, -0.4928, -0.1743, -0.0537],\n",
       "         [ 0.0321,  0.0916, -0.5275, -0.1759, -0.0503],\n",
       "         [ 0.0353,  0.0950, -0.5578, -0.1764, -0.0453],\n",
       "         [ 0.0384,  0.0966, -0.5842, -0.1765, -0.0392]],\n",
       "        grad_fn=<ThAddBackward>),\n",
       " tensor([[-0.2768,  0.0471, -0.0074, -0.1648,  0.1985],\n",
       "         [-0.4023,  0.0643,  0.0055, -0.2424,  0.2919],\n",
       "         [-0.5125,  0.0802,  0.0173, -0.3078,  0.3777],\n",
       "         [-0.6059,  0.0951,  0.0270, -0.3615,  0.4549],\n",
       "         [-0.6829,  0.1095,  0.0340, -0.4048,  0.5235],\n",
       "         [-0.7452,  0.1234,  0.0379, -0.4392,  0.5841],\n",
       "         [-0.7948,  0.1369,  0.0386, -0.4664,  0.6373],\n",
       "         [-0.8339,  0.1502,  0.0360, -0.4875,  0.6840],\n",
       "         [-0.8644,  0.1631,  0.0305, -0.5037,  0.7250],\n",
       "         [-0.8881,  0.1757,  0.0223, -0.5158,  0.7608]],\n",
       "        grad_fn=<ThAddBackward>))"
      ]
     },
     "execution_count": 950,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class MyGRU(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, nonlinearity=torch.tanh, **kwargs):\n",
    "        super().__init__()\n",
    "        self.hidden_size = hidden_size\n",
    "        self.w_z = self.xavier_init([hidden_size, hidden_size])\n",
    "        self.w_r = self.xavier_init([hidden_size, hidden_size])\n",
    "        self.w_h = self.xavier_init([hidden_size, hidden_size])\n",
    "        self.w_y = self.xavier_init([hidden_size, hidden_size])\n",
    "        \n",
    "        self.u_z = self.xavier_init([input_size, hidden_size])\n",
    "        self.u_r = self.xavier_init([input_size, hidden_size])\n",
    "        self.u_h = self.xavier_init([input_size, hidden_size])\n",
    "        \n",
    "        self.b_z = nn.Parameter(torch.zeros([1, hidden_size], requires_grad=True))\n",
    "        self.b_r = nn.Parameter(torch.zeros([1, hidden_size], requires_grad=True))\n",
    "        self.b_h = nn.Parameter(torch.zeros([1, hidden_size], requires_grad=True))\n",
    "        self.b_y = nn.Parameter(torch.zeros([1, hidden_size], requires_grad=True))\n",
    "\n",
    "        self.nonlinearity = nonlinearity\n",
    "    \n",
    "    def forward(self, x):\n",
    "        seq_size = x.size(0)\n",
    "        self.h = torch.zeros([1, self.hidden_size], requires_grad=True)\n",
    "        \n",
    "        for i in range(seq_size):\n",
    "            z = torch.sigmoid(x[i]*self.u_z + self.h@self.w_z + self.b_z)\n",
    "            r = torch.sigmoid(x[i]*self.u_r + self.h@self.w_r + self.b_r)\n",
    "            h_new = torch.tanh(x[i]*self.u_h + (r*self.h)@self.w_h + self.b_h)\n",
    "            \n",
    "            self.h = (1 - z) * self.h + z * h_new\n",
    "            self.o = self.h@self.w_y + self.b_y\n",
    "            \n",
    "        return self.o, self.h\n",
    "    \n",
    "    def xavier_init(self, size):\n",
    "        return nn.Parameter(torch.autograd.Variable(\n",
    "            nn.init.xavier_normal_(torch.empty(*size)),\n",
    "            requires_grad=True\n",
    "        ))\n",
    "n = MyGRU(1, 5)\n",
    "n(data[0:10, :].permute(1, 0).view(sq, -1, 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1026,
   "metadata": {},
   "outputs": [],
   "source": [
    "sq = 4\n",
    "def create_dataset(nums, y, batch_size=sq):\n",
    "    batches = []\n",
    "    targets = []\n",
    "    for i in range(0+batch_size, len(nums)):\n",
    "        batch = nums[i-batch_size:i]\n",
    "        target = y[i]\n",
    "        batches.append(batch)\n",
    "        targets.append(target)\n",
    "    return batches, targets\n",
    "\n",
    "data, targets = create_dataset(x, y, batch_size=sq)\n",
    "data = np.asarray(data)\n",
    "targets = np.asarray(targets)\n",
    "data = torch.autograd.Variable(torch.Tensor(data), requires_grad=False)\n",
    "targets = torch.autograd.Variable(torch.Tensor(targets), requires_grad=False)\n",
    "\n",
    "class MyModel(torch.nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.gru = MyGRU(1, 1, num_layers=1, batch_first=False)\n",
    "    def forward(self, input):\n",
    "        input = input.permute(1, 0).view(sq, -1, 1)\n",
    "        output, hidden = self.gru(input)\n",
    "        \n",
    "        return hidden\n",
    "        \n",
    "model = MyModel()\n",
    "optim = torch.optim.RMSprop(model.parameters(), lr=0.01)\n",
    "criterion = torch.nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1027,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 0.875205397605896\n",
      "100 0.3248729407787323\n",
      "200 0.27816644310951233\n",
      "300 0.2403561770915985\n",
      "400 0.21625037491321564\n",
      "500 0.20399761199951172\n",
      "600 0.1976916491985321\n",
      "700 0.19357292354106903\n",
      "800 0.19048726558685303\n",
      "900 0.180766299366951\n"
     ]
    }
   ],
   "source": [
    "for i in range(1000):\n",
    "    optim.zero_grad()\n",
    "    loss = criterion(model(data), targets.view(-1, 1))\n",
    "    if i % 100 == 0:\n",
    "        print(i, loss.item())\n",
    "    loss.backward()\n",
    "    optim.step()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1028,
   "metadata": {},
   "outputs": [],
   "source": [
    "for param_group in optim.param_groups:\n",
    "    param_group['lr'] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1029,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(96, torch.Size([96]))"
      ]
     },
     "execution_count": 1029,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds = model(data).view(-1)\n",
    "len(preds), preds.size()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1030,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD8CAYAAABzTgP2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBYMxxpiT/H+S1iS5AT+D8QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.arange(len(preds)), preds.view(-1).data.numpy())\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 944,
   "metadata": {},
   "outputs": [],
   "source": [
    "a,b,c = [], [], []\n",
    "for i, batch in enumerate(data):\n",
    "    a.append(batch[-1] + 0.1)\n",
    "    b.append(model(batch.view(1, -1)).view(-1))\n",
    "    # print(model(batch.view(1, 1, 8)))\n",
    "    c.append(targets[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 945,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD8CAYAAABzTgP2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(a, b, label=\"pred\")\n",
    "plt.plot(a, c, label=\"target\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}